How the data is collected
Every finding comes from the same method: the same questions, asked of the same AI assistants, on a fixed schedule, with every answer kept.
What is asked
The questions a buyer asks at each stage of buying: learning about a category, drawing up a shortlist, checking the options and choosing. They are written for one sector and one market.
Which AI assistants
Up to six: ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek. Each is asked the same questions, so their answers can be compared.
How often
Every week or every month, the same questions each time, so a change in the answers can be seen rather than guessed.
What is kept
Every answer in full, the brands it names in the order named, and the pages it cites. A finding's chart is frozen at the reading it was read at.
How a test is judged
Each test states the reading that would prove it wrong, and that threshold is set before the data arrives. If it is changed later, both dates are shown.
The caution
One sector, one market and the AI assistants named, at one reading: a later reading can change a finding. Independent analysis of public data. Not financial advice or a product recommendation.
Run a test on your own brand and market.
Run a test on your own brand